Partially observable Markov decision process-based MAC-layer sensing optimisation for cognitive radios exploiting rateless-coded spectrum aggregation
Xiaohong Wang, Wei Chen, Zhanghua Cao · IET Communications · 2012
Cognitive radio (CR) provides a promising solution to the spectrum scarcity problem by implementing opportunistic spectrum access over the licensed spectrum. However, spectrum holes are discontinuous in frequency and time, resulting in a challenge to CR transmissions. Fortunately, rateless codes can be utilised to exploit these distributed spectrum opportunities in an aggregate way. In such system, how to conduct the sensing and transmission is a key problem that affects the system performance. Therefore in this study, the authors propose a rateless-coded transmission protocol in a multi-channel CR system, addressing the media access control (MAC) layer sensing issues. Specifically, how many channels and which ones should be sensed in each time slot. Owing to the dynamics of channel availability, stochastic control is a necessity. Therefore the authors analyse the average throughput and formulate an optimisation problem to find the optimal sensing policy based on the theory of partially observable Markov decision process (POMDP). The myopic sensing policy is also studied owing to intractable computation complexity of a general POMPD. Moreover, the authors propose a heuristic policy with comparable performance and low complexity. Simulation results will show that the heuristic policy is superior to the static policy and has almost the same performance as the myopic policy.